MNRAS 461, 1841–1849 (2016) doi:10.

1093/mnras/stw1049
Advance Access publication 2016 May 3

An upper boundary in the mass-metallicity plane of exo-Neptunes

Bastien Courcol,‹ François Bouchy and Magali Deleuil

Aix Marseille University, CNRS, Laboratoire d’Astrophysique de Marseille UMR 7326, F-13388 Marseille cedex 13, France

Downloaded from https://academic.oup.com/mnras/article-abstract/461/2/1841/2608353 by The Open University user on 23 April 2020

Accepted 2016 April 28. Received 2016 April 28; in original form 2016 February 11

ABSTRACT
With the progress of detection techniques, the number of low-mass and small-size exoplanets
is increasing rapidly. However their characteristics and formation mechanisms are not yet
fully understood. The metallicity of the host star is a critical parameter in such processes and
can impact the occurrence rate or physical properties of these planets. While a frequency–
metallicity correlation has been found for giant planets, this is still an ongoing debate for their
smaller counterparts. Using the published parameters of a sample of 157 exoplanets lighter
than 40 M⊕ , we explore the mass-metallicity space of Neptunes and super-Earths. We show
the existence of a maximal mass that increases with metallicity, that also depends on the period
of these planets. This seems to favour in situ formation or alternatively a metallicity-driven
migration mechanism. It also suggests that the frequency of Neptunes (between 10 and 40 M⊕ )
is, like giant planets, correlated with the host star metallicity, whereas no correlation is found
for super-Earths (<10 M⊕ ).
Key words: methods: statistical – planetary systems.

We took advantage of the increasing number of low-mass exo-

1 I N T RO D U C T I O N
Thanks to continuous improvements of detection techniques, the
list of known exoplanets has exponentially increased these 20 past
years. More specifically a new population of low-mass or small-size
exoplanets has emerged both from the high-precision radial velocity
surveys (Howard et al. 2009; Mayor et al. 2009) and high-precision
photometric surveys (Baglin 2003; Borucki & Koch 2011). Arbi-
trarily this population usually distinguishes Neptune like objects,
with a mass from 10 to 40 M⊕ and a radius from 2 to 6 R⊕ , and
super-Earths, with a mass from 2 to 10 M⊕ and a radius smaller
than 2 R⊕ . To investigate the properties of the different planet pop-
ulations in the low-mass regime, and explore possible correlation
between their physical properties, it is however required to get a
large sample of planets with parameters accurately determined.
Among the properties of exoplanet host stars, the metallicity
was early identified as a key element for giant gaseous planets. It
has been well established that the occurrence rate of giant planets
increases with metallicity, e.g. Gonzalez (1997); Laws et al. (2003);
Santos et al. (2005); Sousa et al. (2008). Nevertheless, for low-
mass exoplanets such a correlation was not observed (.e.g. Sousa
et al. (2008); Ghezzi et al. (2010); Mayor et al. (2011); Sousa et al.
(2011)). Jenkins et al. (2013) claim the existence of a minimal mass
for super-Earth objects that increases with host star metallicity.
More recently Wang & Fischer (2015), based on Kepler results,
pointed out a universal correlation between the occurrence rate and
the host star metallicity, which is weaker for terrestrial planets.
planets published so far to explore and better quantify the possible
correlation between their mass and the metallicity of their host
star. Section 2 presents the sample of low-mass planets we used.
In Section 3, we study the mass-metallicity diagram and show the
existence of a upper boundary in the mass-metallicity plane as well
as investigate its correlation with the period. Section 4 details the
impact of this mass-metallicity trend on the frequency of low-mass
planets. We discuss our findings in Section 5 and finally present our
conclusions in the Section 6.
2 THE SAMPLE
The sample contains all the known low-mass exoplanets (Msin(i) <
40M⊕ ) with a precision on the measured mass better than
20 per cent and a precision on the metallicity index [Fe/H] better
than 0.2 dex. It was built on the basis of the main websites exo-
planets catalogues: the NASA Exoplanet Archive,1 exoplanets.org
and exoplanet.eu. The metallicities, planetary masses and other pa-
rameters were carefully cross-checked between catalogues. We also
included 44 planets from Mayor et al. (2011) as they are already
present in the exoplanet.eu data base, updated by the recent resub-
mitted version (private communication). This new set of low-mass
planets increases the size of the sample by 27 per cent but do not
change nor impact our results.
The final list with references is displayed in Table 1. It contains
157 planets with masses and metallicities ranging from 1.13 M⊕ to

 E-mail: bastien.courcol@lam.fr 1 http://exoplanetarchive.ipac.caltech.edu/index.html


C 2016 The Authors

Published by Oxford University Press on behalf of the Royal Astronomical Society

1842 B. Courcol, F. Bouchy and M. Deleuil
Table 1. Annex: parameters of the sample.

Name Mass [M⊕ ] [Fe/H] Period [d] References

Kepler-78 b 1.86 ± 0.30 −0.14 ± 0.08 0.35 Pepe et al. (2013), Sanchis-Ojeda et al. (2013)
55 Cnc e 8.32 ± 0.39 0.33 ± 0.07 0.74 Endl et al. (2012), Santos et al. (2013)
Kepler-10 b 3.33 ± 0.49 −0.15 ± 0.04 0.84 Dumusque et al. (2014), Santos et al. (2013)
GJ 1214 b 6.47 ± 1.00 0.01 ± 0.20 1.58 Carter et al. (2011), Santos et al. (2013)
GJ 876 d 5.85 ± 0.39 0.15 ± 0.10 1.94 Rivera et al. (2010), Santos et al. (2013)
GJ 436 b 23.06 ± 1.01 0.01 ± 0.20 2.64 Maness et al. (2007), Santos et al. (2013)
7.05 ± 0.87 −0.04 ± 0.20

Downloaded from https://academic.oup.com/mnras/article-abstract/461/2/1841/2608353 by The Open University user on 23 April 2020

GJ 3634 b 2.65 Bonfils et al. (2011), Santos et al. (2013)
GJ 581 e 1.95 ± 0.22 0.21 ± 0.10 3.15 Forveille et al. (2011), Santos et al. (2013)
HATS-7 b 38.00 ± 3.80 0.25 ± 0.08 3.18 Bakos et al. (2015)
Kepler-4 b 24.50 ± 3.80 0.17 ± 0.06 3.20 Borucki et al. (2010), Santos et al. (2013)
alpha Cen B b 1.13 ± 0.10 0.16 ± 0.04 3.24 Dumusque et al. (2012), Santos et al. (2013)
61 Vir b 5.10 ± 0.60 0.01 ± 0.05 4.20 Vogt et al. (2010), Santos et al. (2013)
61 Vir c 18.20 ± 1.10 0.01 ± 0.05 38.00 Vogt et al. (2010), Santos et al. (2013)
61 Vir d 22.90 ± 2.60 0.01 ± 0.05 123.00 Vogt et al. (2010), Santos et al. (2013)
BD -08 2823 b 14.60 ± 1.01 0.00 ± 0.08 5.60 Hébrard et al. (2010), Santos et al. (2013)
BD-061339 b 6.30 ± 0.80 −0.14 ± 0.17 3.87 Tuomi (2014), Santos et al. (2013)
CoRoT-7 c 13.56 ± 1.08 0.02 ± 0.02 3.70 Haywood et al. (2014), Santos et al. (2013)
GJ 15 A b 5.34 ± 0.76 −0.32 ± 0.17 11.44 Howard et al. (2014)
GJ 160.2 b 10.20 ± 2.00 0.00 ± 0.15 5.24 Tuomi (2014), Soubiran et al. (2010)
GJ 163 b 10.77 ± 0.85 −0.02 ± 0.20 8.63 Bonfils et al. (2013), Santos et al. (2013)
GJ 163 c 6.85 ± 0.99 −0.02 ± 0.20 25.63 Bonfils et al. (2013), Santos et al. (2013)
GJ 163 d 29.43 ± 4.05 −0.02 ± 0.20 603.95 Bonfils et al. (2013), Santos et al. (2013)
GJ 176 b 8.40 ± 1.00 −0.01 ± 0.10 8.80 Forveille et al. (2009), Santos et al. (2013)
GJ 3293 b 24.00 ± 1.70 0.02 ± 0.09 30.60 Astudillo-Defru et al. (2015)
GJ 3293 d 22.30 ± 1.70 0.02 ± 0.09 124.00 Astudillo-Defru et al. (2015)
GJ 3470 b 13.90 ± 1.50 0.08 ± 0.10 3.30 Demory et al. (2013), Santos et al. (2013)
GJ 433 b 5.78 ± 0.47 −0.17 ± 0.10 7.37 Delfosse et al. (2013), Santos et al. (2013)
GJ 581 b 15.86 ± 0.72 0.21 ± 0.10 5.37 Forveille et al. (2011), Santos et al. (2013)
GJ 581 c 5.33 ± 0.38 0.21 ± 0.10 12.92 Forveille et al. (2011), Santos et al. (2013)
GJ 667 C b 5.56 ± 0.34 −0.53 ± 0.10 7.20 Robertson & Mahadevan (2014), Santos et al. (2013)
GJ 667 C c 4.15 ± 0.68 −0.53 ± 0.10 28.10 Robertson & Mahadevan (2014), Santos et al. (2013)
GJ 667C d 5.10 ± 0.60 −0.53 ± 0.10 91.61 Anglada-Escudé et al. (2013), Santos et al. (2013)
GJ 667C e 2.70 ± 0.50 −0.53 ± 0.10 62.24 Anglada-Escudé et al. (2013), Santos et al. (2013)
GJ 667C f 3.80 ± 0.40 −0.53 ± 0.10 28.14 Anglada-Escudé et al. (2013), Santos et al. (2013)
GJ 667C g 4.60 ± 0.80 −0.53 ± 0.10 256.20 Anglada-Escudé et al. (2013), Santos et al. (2013)
GJ 674 b 11.09 ± 0.24 −0.25 ± 0.10 4.70 Bonfils et al. (2007), Santos et al. (2013)
GJ 676A d 4.40 ± 0.70 0.08 ± 0.20 3.60 Anglada-Escudé & Tuomi (2012), Santos et al. (2013)
GJ 676A e 11.50 ± 1.50 0.08 ± 0.20 35.37 Anglada-Escudé & Tuomi (2012), Santos et al. (2013)
GJ 832 c 5.00 ± 1.00 −0.19 ± 0.10 35.70 Wittenmyer et al. (2014), Santos et al. (2013)
GJ 876 e 12.47 ± 1.62 0.15 ± 0.10 124.26 Rivera et al. (2010), Santos et al. (2013)
Gl 687 b 18.00 ± 2.00 −0.09 ± 0.15 38.10 Burt et al. (2014)
Gl 785 b 21.60 ± 2.00 0.08 ± 0.03 74.40 Howard et al. (2011b)
HAT-P-11 b 26.22 ± 2.86 0.26 ± 0.08 4.89 Bakos et al. (2010), Santos et al. (2013)
HAT-P-26 b 18.70 ± 2.20 0.01 ± 0.04 4.23 Hartman et al. (2011), Santos et al. (2013)
HD 10180 c 13.19 ± 0.62 0.08 ± 0.01 5.76 Lovis et al. (2011), Santos et al. (2013)
HD 10180 d 11.97 ± 0.77 0.08 ± 0.01 16.36 Lovis et al. (2011), Santos et al. (2013)
HD 10180 e 25.36 ± 1.37 0.08 ± 0.01 49.75 Lovis et al. (2011), Santos et al. (2013)
HD 10180 f 23.62 ± 1.66 0.08 ± 0.01 122.72 Lovis et al. (2011), Santos et al. (2013)
HD 10180 g 21.41 ± 2.97 0.08 ± 0.01 602.00 Lovis et al. (2011), Santos et al. (2013)
HD 102365 b 16.20 ± 2.58 −0.29 ± 0.02 122.10 Tinney et al. (2011), Santos et al. (2013)
HD 103197 b 31.22 ± 1.90 0.22 ± 0.04 47.84 Mordasini et al. (2011), Santos et al. (2013)
HD 109271 b 17.00 ± 1.00 0.10 ± 0.01 7.90 Lo Curto et al. (2013), Santos et al. (2013)
HD 109271 c 24.00 ± 2.00 0.10 ± 0.01 30.90 Lo Curto et al. (2013), Santos et al. (2013)
HD 11964 c 24.49 ± 3.51 0.14 ± 0.05 37.91 Wright et al. (2009), Santos et al. (2013)
HD 125595 b 13.25 ± 1.37 0.10 ± 0.14 9.67 Ségransan et al. (2011), Santos et al. (2013)
HD 125612 c 18.45 ± 3.28 0.24 ± 0.01 4.15 Lo Curto et al. (2010), Santos et al. (2013)
HD 134060 b 11.17 ± 0.66 0.14 ± 0.01 38.00 Mayor et al. (2011), Santos et al. (2013)
HD 134606 b 2.37 ± 0.28 0.27 ± 0.02 4.30 Mayor et al. (2011), Santos et al. (2013)
HD 134606 c 9.26 ± 0.42 0.27 ± 0.02 12.10 Mayor et al. (2011), Santos et al. (2013)
HD 134606 d 5.20 ± 0.58 0.27 ± 0.02 26.90 Mayor et al. (2011), Santos et al. (2013)
HD 134606 e 10.70 ± 0.76 0.27 ± 0.02 58.80 Mayor et al. (2011), Santos et al. (2013)
HD 134606 f 6.90 ± 1.20 0.27 ± 0.02 147.50 Mayor et al. (2011), Santos et al. (2013)
HD 136352 b 5.28 ± 0.62 −0.34 ± 0.01 11.56 Mayor et al. (2011), Santos et al. (2013)
HD 136352 c 11.38 ± 0.10 −0.34 ± 0.01 27.60 Mayor et al. (2011), Santos et al. (2013)

MNRAS 461, 1841–1849 (2016)

 An upper boundary in the mass-metallicity plane of exo-Neptunes 1843
Table 1 – continued

Name Mass [M⊕ ] [Fe/H] Period [d] References

HD 136352 d 9.59 ± 1.86 −0.34 ± 0.01 106.70 Mayor et al. (2011), Santos et al. (2013)
HD 13808 b 10.33 ± 0.92 −0.21 ± 0.02 14.20 Mayor et al. (2011), Santos et al. (2013)
HD 13808 c 11.55 ± 1.62 −0.21 ± 0.02 53.80 Mayor et al. (2011), Santos et al. (2013)
HD 1461 b 6.44 ± 0.61 0.19 ± 0.01 13.50 Dı́az et al. (2016), Santos et al. (2013)
HD 1461 c 5.92 ± 0.76 0.19 ± 0.01 13.50 Dı́az et al. (2016), Santos et al. (2013)
HD 154088 b 6.15 ± 0.86 0.28 ± 0.03 18.60 Mayor et al. (2011), Santos et al. (2013)
4.15 ± 0.59 −0.04 ± 0.05

Downloaded from https://academic.oup.com/mnras/article-abstract/461/2/1841/2608353 by The Open University user on 23 April 2020

HD 156668 b 4.65 Howard et al. (2011a), Santos et al. (2013)
HD 157172 b 38.10 ± 2.60 0.11 ± 0.02 104.80 Mayor et al. (2011), Santos et al. (2013)
HD 16417 b 21.28 ± 1.89 0.13 ± 0.01 17.24 O’Toole et al. (2009a), Santos et al. (2013)
HD 164595 b 16.14 ± 2.72 −0.04 ± 0.08 40.00 Courcol et al. (2015), Porto de Mello et al. (2014)
HD 179079 b 27.50 ± 2.50 0.27 ± 0.02 14.48 Valenti et al. (2009), Santos et al. (2013)
HD 181433 b 7.54 ± 0.68 0.36 ± 0.18 9.37 Bouchy et al. (2009), Santos et al. (2013)
HD 189567 b 8.46 ± 0.59 −0.24 ± 0.01 14.30 Mayor et al. (2011), Santos et al. (2013)
HD 189567 c 7.23 ± 0.85 −0.24 ± 0.01 33.62 Mayor et al. (2011), Santos et al. (2013)
HD 189567 d 7.40 ± 1.40 −0.24 ± 0.01 61.72 Mayor et al. (2011), Santos et al. (2013)
HD 190360 c 18.74 ± 2.12 0.24 ± 0.05 17.11 Wright et al. (2009), Santos et al. (2013)
HD 192310 b 16.90 ± 0.90 −0.03 ± 0.04 74.72 Pepe et al. (2011), Santos et al. (2013)
HD 20003 b 12.00 ± 0.97 0.04 ± 0.02 11.90 Mayor et al. (2011), Santos et al. (2013)
HD 20003 c 13.42 ± 1.28 0.04 ± 0.02 33.80 Mayor et al. (2011), Santos et al. (2013)
HD 204313 c 17.6 ± 1.7 0.18 ± 0.02 34.90 Dı́az et al. (2016), Santos et al. (2013)
HD 20781 b 3.83 ± 0.73 −0.11 ± 0.02 5.30 Mayor et al. (2011), Santos et al. (2013)
HD 20781 c 5.51 ± 0.47 −0.11 ± 0.02 13.90 Mayor et al. (2011), Santos et al. (2013)
HD 20781 d 10.63 ± 0.63 −0.11 ± 0.02 29.20 Mayor et al. (2011), Santos et al. (2013)
HD 20781 e 15.30 ± 0.81 −0.11 ± 0.02 85.40 Mayor et al. (2011), Santos et al. (2013)
HD 20794 b 2.70 ± 0.31 −0.40 ± 0.01 18.32 Pepe et al. (2011), Santos et al. (2013)
HD 20794 c 2.36 ± 0.43 −0.40 ± 0.01 40.11 Pepe et al. (2011), Santos et al. (2013)
HD 20794 d 4.70 ± 0.57 −0.40 ± 0.01 90.31 Pepe et al. (2011), Santos et al. (2013)
HD 215152 b 2.78 ± 0.47 −0.08 ± 0.02 7.28 Mayor et al. (2011), Santos et al. (2013)
HD 215152 c 10.00 ± 0.48 −0.08 ± 0.02 10.87 Mayor et al. (2011), Santos et al. (2013)
HD 215456 b 32.21 ± 2.92 −0.09 ± 0.01 193.00 Mayor et al. (2011), Santos et al. (2013)
HD 215497 b 6.63 ± 0.79 0.25 ± 0.05 3.93 Lo Curto et al. (2010), Santos et al. (2013)
HD 21693 b 10.22 ± 1.46 0.00 ± 0.02 22.70 Mayor et al. (2011), Santos et al. (2013)
HD 21693 c 20.57 ± 1.80 0.00 ± 0.02 53.90 Mayor et al. (2011), Santos et al. (2013)
HD 219828 b 19.77 ± 1.56 0.19 ± 0.03 3.83 Melo et al. (2007), Santos et al. (2013)
HD 31527 b 11.55 ± 0.80 −0.17 ± 0.01 16.50 Mayor et al. (2011), Santos et al. (2013)
HD 31527 c 15.82 ± 1.10 −0.17 ± 0.01 51.30 Mayor et al. (2011), Santos et al. (2013)
HD 31527 d 16.50 ± 3.00 −0.17 ± 0.01 274.50 Mayor et al. (2011), Santos et al. (2013)
HD 38858 b 12.43 ± 1.70 −0.22 ± 0.01 198.00 Mayor et al. (2011), Santos et al. (2013)
HD 39194 b 3.72 ± 0.33 −0.61 ± 0.02 5.63 Mayor et al. (2011), Santos et al. (2013)
HD 39194 c 5.94 ± 0.47 −0.61 ± 0.02 14.03 Mayor et al. (2011), Santos et al. (2013)
HD 39194 d 5.14 ± 0.66 −0.61 ± 0.02 33.90 Mayor et al. (2011), Santos et al. (2013)
HD 40307 b 3.81 ± 0.3 −0.36 ± 0.02 4.31 Dı́az et al. (2016), Santos et al. (2013)
HD 40307 c 6.43 ± 0.44 −0.36 ± 0.02 9.62 Dı́az et al. (2016), Santos et al. (2013)
HD 40307 d 8.74 ± 0.58 −0.36 ± 0.02 20.42 Dı́az et al. (2016), Santos et al. (2013)
HD 40307 e 3.52 ± 0.13 −0.36 ± 0.02 34.62 Tuomi et al. (2013), Santos et al. (2013)
HD 40307 f 3.63 ± 0.6 −0.36 ± 0.02 51.56 Dı́az et al. (2016), Santos et al. (2013)
HD 40307 g 7.10 ± 0.90 −0.36 ± 0.02 197.80 Tuomi et al. (2013), Santos et al. (2013)
HD 4308 b 13.00 ± 1.40 −0.34 ± 0.01 15.56 O’Toole et al. (2009b), Santos et al. (2013)
HD 45184 b 11.32 ± 0.83 0.04 ± 0.01 5.90 Mayor et al. (2011), Santos et al. (2013)
HD 45184 c 8.98 ± 1.13 0.04 ± 0.01 13.13 Mayor et al. (2011), Santos et al. (2013)
HD 47186 b 22.63 ± 0.88 0.23 ± 0.02 4.08 Bouchy et al. (2009), Santos et al. (2013)
HD 49674 b 32.28 ± 2.61 0.33 ± 0.06 4.95 Butler et al. (2006), Santos et al. (2013)
HD 51608 b 13.14 ± 0.98 −0.07 ± 0.01 14.10 Mayor et al. (2011), Santos et al. (2013)
HD 51608 c 17.97 ± 2.61 −0.07 ± 0.01 95.42 Mayor et al. (2011), Santos et al. (2013)
HD 69830 b 10.06 ± 0.55 −0.06 ± 0.02 8.67 Lovis et al. (2006), Santos et al. (2013)
HD 69830 c 11.69 ± 0.81 −0.06 ± 0.02 31.56 Lovis et al. (2006), Santos et al. (2013)
HD 69830 d 17.90 ± 1.66 −0.06 ± 0.02 197.00 Lovis et al. (2006), Santos et al. (2013)
HD 7924 b 8.68 ± 0.52 −0.22 ± 0.04 5.40 Fulton et al. (2015), Santos et al. (2013)
HD 7924 c 7.86 ± 0.72 −0.22 ± 0.04 15.30 Fulton et al. (2015), Santos et al. (2013)
HD 7924 d 6.44 ± 0.79 −0.22 ± 0.04 24.45 Fulton et al. (2015), Santos et al. (2013)
HD 85512 b 3.62 ± 0.44 −0.26 ± 0.14 58.43 Pepe et al. (2011), Santos et al. (2013)
HD 90156 b 17.97 ± 1.49 −0.24 ± 0.01 49.77 Mordasini et al. (2011), Santos et al. (2013)
HD 93385 b 4.02 ± 0.48 0.02 ± 0.01 7.34 Mayor et al. (2011), Santos et al. (2013)
HD 93385 c 7.20 ± 0.58 0.02 ± 0.01 13.18 Mayor et al. (2011), Santos et al. (2013)

MNRAS 461, 1841–1849 (2016)

1844 B. Courcol, F. Bouchy and M. Deleuil
Table 1 – continued

Name Mass [M⊕ ] [Fe/H] Period [d] References

HD 93385 d 7.78 ± 0.87 0.02 ± 0.01 45.84 Mayor et al. (2011), Santos et al. (2013)
HD 96700 b 9.08 ± 0.41 −0.18 ± 0.01 8.13 Mayor et al. (2011), Santos et al. (2013)
HD 96700 c 3.22 ± 0.56 −0.18 ± 0.01 19.90 Mayor et al. (2011), Santos et al. (2013)
HD 96700 d 12.25 ± 0.98 −0.18 ± 0.01 103.22 Mayor et al. (2011), Santos et al. (2013)
HD 97658 b 7.55 ± 0.80 −0.35 ± 0.02 9.49 Van Grootel et al. (2014), Santos et al. (2013)
HD 99492 b 33.75 ± 3.72 0.24 ± 0.12 17.04 Butler et al. (2006), Santos et al. (2013)
11.82 ± 1.33 −0.12 ± 0.03

Downloaded from https://academic.oup.com/mnras/article-abstract/461/2/1841/2608353 by The Open University user on 23 April 2020

HIP 116454 b 9.12 Vanderburg et al. (2015), Santos et al. (2013)
HIP 57274 b 11.62 ± 1.31 0.01 ± 0.06 8.14 Fischer et al. (2012), Santos et al. (2013)
Kepler-10 c 17.20 ± 1.90 −0.15 ± 0.04 45.30 Dumusque et al. (2014), Santos et al. (2013)
Kepler-11 d 7.30 ± 1.00 0.00 ± 0.10 22.70 Lissauer et al. (2013), Santos et al. (2013)
Kepler-18 c 18.40 ± 2.70 0.20 ± 0.04 7.64 Hadden & Lithwick (2014), Santos et al. (2013)
Kepler-18 d 15.70 ± 2.00 0.20 ± 0.04 14.86 Hadden & Lithwick (2014), Santos et al. (2013)
Kepler-307 b 3.10 ± 0.60 0.16 ± 0.15 10.42 Xie (2014), MAST catalogue
Kepler-36 b 4.45 ± 0.33 −0.20 ± 0.06 13.84 Carter et al. (2012), Santos et al. (2013)
Kepler-36 c 8.08 ± 0.60 −0.20 ± 0.06 16.24 Carter et al. (2012), Santos et al. (2013)
Kepler-48 c 14.61 ± 2.30 0.17 ± 0.07 9.67 Marcy et al. (2014)
Kepler-51 c 4.00 ± 0.40 −0.08 ± 0.15 85.30 Masuda (2014), MAST catalogue
Kepler-56 b 22.10 ± 3.70 0.20 ± 0.16 10.50 Huber et al. (2013)
Kepler-89 e 13.00 ± 2.50 −0.01 ± 0.04 54.30 Masuda et al. (2013), Hirano et al. (2012)
Kepler-93 b 4.02 ± 0.68 −0.18 ± 0.10 4.73 Dressing et al. (2015)
KOI-620.02 7.60 ± 1.10 −0.08 ± 0.15 130.20 Masuda (2014), MAST catalogue
mu Ara c 10.50 ± 0.50 0.32 ± 0.04 9.63 Pepe et al. (2007), Santos et al. (2013)
mu Ara d 10.99 ± 0.63 0.32 ± 0.04 9.64 Pepe et al. (2007), Santos et al. (2013)
Kapteyn’s c 7.00 ± 1.10 −0.89 ± 0.15 121.54 Anglada-Escudé et al. (2014)
HD 219134 b 4.46 ± 0.47 0.11 ± 0.04 3.09 Motalebi et al. (2015)
HD 219134 d 8.67 ± 1.14 0.11 ± 0.04 46.78 Motalebi et al. (2015)
HD 219134 f 8.90 ± 1.00 0.11 ± 0.04 22.80 Vogt et al. (2015), Motalebi et al. (2015)
HD 219134 g 11.00 ± 1.00 0.11 ± 0.04 94.20 Vogt et al. (2015), Motalebi et al. (2015)
HD 175607 b 8.98 ± 1.1 −0.62 ± 0.01 29.01 Mortier et al. (2016)

38.1 M⊕ and from −0.89 to 0.39 dex, respectively. The stellar type
of the host stars range from M to F. We note that 88 per cent of the
planets have periods less than 100 d. 25 planets were detected in
transit, including some planets with masses determined by transit
timing variations (TTVs), emphasizing the small overlap between
radial velocity and transit surveys.
Due to the multiplicity of the sources, there is no uniform metal-
licity determination method. Moreover, in some cases, the results
of the different methods wildly disagree, with differences that can
reach ∼0.3 dex (Johnson & Apps 2009; Neves et al. 2012). To mit-
igate that effect, that could induce biases, we used the metallicity
values from SWEET-Cat (Santos et al. 2013) whenever possible.
This catalogue aims at determining atmospheric parameters of ex-
oplanet host stars in the most uniform way possible using the same
methodology as well as compile values in the literature in a way that
optimizes the uniformity, making them more suitable for statistical
Figure 1. Planetary mass/host star metallicity diagram for all known planets
studies of stars with planets. In the present case 134 of our 157
lighter than 40 M⊕ , with M/M < 0.2 and [Fe/H] < 0.2 dex. Red dots are
planets, in 80 of the 97 systems, are present in the catalogue. We transiting planets. Some errors bars do not appear either because errors were
note that the remaining planets are quite uniformly distributed in not provided or because it is below the size of the dot. The dashed green
the parameters space and should not introduce any significant bias. line is the computed boundary, and the dashed blue line its approximation
described in equation (1) (see text).
3 T H E M A S S - M E TA L L I C I T Y D I AG R A M
that increases with metallicity i.e. an upper boundary, or that there
is an exclusion zone in the upper left part of the Fig. 1, that is high
3.1 On the existence of an exclusion zone
masses and low metallicity.
The planetary minimal mass/host star metallicity diagram is pre- This type of dependence between two parameters is unusual and
sented in Fig. 1. While a connection between mass and metallicity might be interpreted at first glance by a bias in the sample, as
is obvious, to describe it as a correlation would be misleading as the the dispersion of the mass increases with metallicity. First, biases
mass does not necessarily increase with the host star metallicity. A related to the stellar type should be reviewed. Low-mass planets
more adequate description would be that there is a maximum mass are more easily detected around M dwarves because of their lower

MNRAS 461, 1841–1849 (2016)

 An upper boundary in the mass-metallicity plane of exo-Neptunes 1845
masses. Additionally, the precise determination of the metallicity
of these stars is more difficult. However, when removing all the
M dwarves from the data, the shape of the exclusion zone remains
the same. More generally, as explained in Section 2, the use of
SWEET-Cat should prevent any bias caused by different metallicity
determination methods.
Furthermore, such a bias cannot be observational in nature, as it
is the low-mass planets that are the hardest to detect that are found
at low metallicity. Small planets orbiting low-metallicity host stars

Downloaded from https://academic.oup.com/mnras/article-abstract/461/2/1841/2608353 by The Open University user on 23 April 2020

are the most difficult to detect, because of the much lower number
of spectral features that can be exploited to obtain a precise radial
velocity measurement. Furthermore, if more massive planets would
have existed in these systems, they should have been detected.
This exclusion region cannot be explained either by a bias in the
angle i between the plane of the system and the line of sight, which
is unknown for most planets in this sample. The distribution of i is a
purely geometrical effect that is not linked to the stellar metallicity.
We also note that the number of transiting planets for which the
true mass is known is small in this mass domain. However, from
a statistical point of view, the use of the minimal mass instead
of the true mass should not significantly change the shape of this
distribution, but only raise it by ∼15 per cent. For this reason, and
for the sake of simplicity, the general term ‘mass’ will be used
hereafter instead of ‘minimal mass’ (or ‘true mass’), except where
a distinction is needed.
The only possible type of bias would then come from differences
in the completeness in period of surveys focusing on either end of
the metallicity range. If the periods probed around low metallicity
stars are significantly shorter than high metallicity stars, it could
explain the trend in Fig. 1 if Neptune-like planets are preferentially
located at longer periods. However that cannot be the case, because
the periods of the planets are not correlated with the metallicity. We
discuss further this point in Section 3.3. Additionally, we could set
up a strict period criterium to ensure a homogeneous completeness
of the sample. If we discard all the planets at periods greater than
10 d (95), the general shape of the mass/metallicity diagram in Fig. 1
does not change.
We finally explored the underlying population distribution in dif-
ferent [Fe/H] bins. To that purpose we divided our sample in three
sub-samples of increasing metallicity with approximately the same
number of planets. Assuming a Poissonnian noise, we checked that
each sub-sample can be phenomenogically described by a single
Gaussian function within the error bars (cf Fig. 2). This is an ad-
ditional evidence that biases are not correlated with [Fe/H] and
therefore could not explain the exclusion zone. We also note that,
as expected, the mean and the dispersion of the Gaussians increase
with the metallicity.

3.2 Determination of the upper boundary

It is possible to define a mass-metallicity boundary separating the
planets from the exclusion zone. To determine its shape we com-
puted the cumulative distribution of planetary masses over a suc-
cession of metallicity bins. To account for the error bars and the
possibility of outliers, each mass is weighted by the inverse of its
precision. The ‘maximum mass’ of the bin is set as the 97 per cent
limit of this cumulative weighted distribution. The extremum bins
(at −0.9 and 0.4 dex) are set to the closest bin value to overcome
boundary effects. The green dashed line on Fig. 1 is the limit derived
with this method with bins centred every 0.1 dex and 0.25 dex wide
(therefore overlapping, to smooth the limit). Modifying the bin size
Figure 2. Mass distributions for metallicity bins of ∼50 planets (blue his-
tograms) and the corresponding Gaussian fits (green curves). The metallicity
range of the bins and the parameters of the Gaussian fits (mean M and dis-
persion σ M ) are, from top to bottom: −0.89 to −0.15 dex, M = 6.4 M⊕ ,
σ M = 4.2 M⊕ ; −0.15 to 0.04 dex, M = 11.6 M⊕ , σ M = 6.2 M⊕ ; −0.04 to
0.39 dex, M = 13.7 M⊕ , σ M = 11.2 M⊕ . The errors bars correspond to a
Poisson noise.

MNRAS 461, 1841–1849 (2016)

1846 B. Courcol, F. Bouchy and M. Deleuil
and spacing can slightly change the shape of the boundary, without
any significant impact.
The main characteristic of the boundary is a monotonous increase
of the maximal mass with [Fe/H]. For metallicities above −0.5, the
trend seems linear. We therefore performed a linear regression on
the boundary to get a simple relation approximating Mmax (in M⊕ )
as a function of [Fe/H] in equation (1).
[Fe/H] > −0.5 : Mmax = 43.3 × [Fe/H] + 29.2M⊕ (1)
For metallicities below −0.5, the boundary is rather flat. However

Downloaded from https://academic.oup.com/mnras/article-abstract/461/2/1841/2608353 by The Open University user on 23 April 2020

the reality of this plateau is questionable as it relies only on one
peculiar planet, Kapteyn’s c (Anglada-Escudé et al. 2014). Its host
is an old sub-M dwarf of the halo, the only one of the sample, which
has a singular metallicity of −0.89 dex with no associated error.
Moreover the orbital period of Kapteyn’s c is among the longest of
the sample: 121.5 d. It is possible that the planetary properties of
such systems are different and that the flat trend is not representative
of the global population.

3.3 Correlation with the period

The unusual nature of this connection between the planetary mass
and the stellar metallicity could be explained by the existence of cor-
relations with other parameters. We therefore investigate a possible
correlation with the orbital period. The Fig. 3, top panel, represents
the mass-metallicity plane, with the logarithm of the period as the
colour scale. We notice that the planets closer to the limit tend to
have longer period compared to those farther away. Moreover, the
few planets that are slightly above the defined boundary all have
periods greater than 100 d. This is also the case of Kapteyn’s c,
which is responsible of the questionable plateau for extremely low
metallicities. This can be best seen in the bottom panel, which rep-
resents log P versus Mmax − Mplanet , i.e. the vertical distance to the Figure 3. Upper panel: distribution of the sample in the mass/metallicity
boundary as defined by the equation (1) (although we did not allow plane with log P as the colour scale. Bottom panel: log P versus Mmax −
Mmax to increase further than 40 M⊕ as it is the limit of our sample). Mplanet , Mmax computed from equation 1. The solid blue line is the linear
When performing a linear regression in the data (the blue dotted regression of the data and the dashed blue lines are the 1σ confidence
line in the bottom panel), we obtain the following relation: interval.

Mmax − Mplanet = −7.15 ± 1.17 × log P + 24.17 ± 1.6M⊕ (2)
For this regression we used an identical weight for all the plan-
ets. Indeed, in this sample, the uncertainties on the mass are not
homogeneously computed. Moreover, other sources of uncertain-
ties should be taken into account in the error on the distance to the
boundary (e.g. the uncertainty introduced by the sin(i), the error on
the position of the limit) that are beyond the scope of this paper.
There is consequently no solid argument to give more weight to
some planets.
The parameters of the linear regression are significantly con-
strained (3.08σ for the slope), although the dispersion of the resid-
uals is quite high. Similarly, the Pearson correlation coefficient of
this data set is of −0.44, but the probability of the no-correlation hy-
pothesis (P-value) is of 8e-9. This means that while the correlation
between log (P) and Mmax − Mplanet is weak, it is very significative.
This result indicates the upper limit decreases for short period
planets. More interestingly, this also suggests that Neptune-like
planets could still exist around metal-poor stars, but at longer
periods.
3.4 On the existence of a lower boundary
Jenkins et al. (2013) also discussed of the possible correlations
between mass and metallicity for low-mass planets. Their study,
based on the exoplanets.org data base as of 2012, focused on a
smaller range of masses (0–19 M⊕ ) and metallicity (−0.5 dex to
0.5 dex). Instead of a correlation, they proposed the existence of
a lower boundary, increasing linearly from 0 M⊕ at −0.2 dex to
9.5 M⊕ at 0.5 dex.
With a sample tripled in size, 121 planets in our study instead of
36 in Jenkins et al. (2013) in the same mass-metallicity range, it is
possible to test this boundary with a better reliability. Our sample is
represented in Fig. 4 in their mass-metallicity range. Eight planets
are found below this boundary (the orange dashed line), three of
them (alpha Cen B b, HD 134606 b, GJ 876 d) at more than 1σ if
we consider conservative errors in [Fe/H] of at least 0.1 dex. This
boundary therefore does not hold up when faced to new detections.
4 I M PAC T O N T H E F R E Q U E N C Y O F S M A L L
PLANETS
The limit between Neptune and super-Earths is difficult to place.
Currently, no clear mass criterium exists to discriminate these two
populations, either observationally or physically motivated. This
can be explained by three reasons. First, degeneracies in planet inte-
riors models make the categorization uncertain for a range of scenar-
ios. Secondly, the existence of transitional planets (mini-Neptunes,
mega-Earths or low density super-Earths) can further scramble the

MNRAS 461, 1841–1849 (2016)

 An upper boundary in the mass-metallicity plane of exo-Neptunes
Figure 4. Mass-metallicity distribution of known small planets. The orange
dashed line is the lower boundary proposed by Jenkins et al. (2013).
Figure 5. Number of the Neptune-mass planets (between 10 and 40 M⊕ )
(green solid line) and super-Earth planets (<10 M⊕ ) (blue dotted line).
Interestingly, the number of planets in each category is similar (80 Neptunes
and 75 super-Earth).
limit to an extent that is currently unknown. And finally, there is a
lack of observational constrains with only 25 small planets with a
relatively well measured density. Consequently we can only choose
an arbitrary mass to define ‘Neptune-mass’ and ‘Super-Earth’ plan-
ets, in this case the widely used 10 M⊕ . However, we can already
perceive different statistical behaviours with the current sample.
Fig. 5, shows the number of ‘Neptune-mass’ planets (between
10 and 40 M⊕ ) as a function of the metallicity (green solid curve).
The shape of this distribution cannot be used as an occurrence rate,
as the planets come from many sources with different selection
biases, methods and detection performances. None the less, it gives
an important clue. One can see that the distribution drops to 0 at
−0.4 dex for the whole sample (green curve), which is coherent
with the shape of the exclusion zone. This implies that at the first
order, the frequency of Neptunes is correlated to metallicity, with no
Neptunes around very sub-metallic stars. Beyond that observation,
it would not be surprising that the frequency of Neptunes do follow
the observed distribution at low metallicities and steadily decrease
to reach 0. Super-Earths (smaller than 10 M⊕ , blue dotted curve in
Fig. 5) on the other hand, are present at all metallicities. We note
that this is still true for all limit masses used to define Neptunes and
super-Earths between 8 and 15 M⊕ .
1847
This result differ from previous studies (Ghezzi et al. 2010; Mayor
et al. 2011; Sousa et al. 2011). In the study of 582 FGK stars of
a HARPS volume limited subsample of Sousa et al. (2011), the
metallicity of Neptunian hosts is rather flat compared to that of the
stars hosting jovians, although they could not achieve statistically
meaningful results due to small numbers. Mayor et al. (2011) present
a flat, metal poor (<0.2 dex besides one exception) distribution for
planets less massive that 30 M⊕ . Finally Ghezzi et al. (2010) also
found a flat relative frequency for Neptune-mass hosts in regard to
Downloaded from https://academic.oup.com/mnras/article-abstract/461/2/1841/2608353 by The Open University user on 23 April 2020
the metallicity, even when adding planets from the literature to their
results. Complementary to these observational results, Mordasini
et al. (2012) reported no correlation between the protoplanetary
disc metallicity and the frequency of Neptunes using formation
models.
Two factor may explain this discrepancy. First, the number of
planets used in our study is much larger and therefore more statisti-
cally reliable. Secondly, the previous studies on this matter seldom
make the distinction between Neptunes and super-Earths, and some-
times define ‘Neptunes’ as any planet with a mass lower than a given
value (25 M⊕ in Ghezzi et al. 2010, 30 M⊕ in Mayor et al. 2011,
0.1 MJup in Sousa et al. 2011). This is important considering that
the distribution of Super-Earths (smaller than 10 M⊕ ) is different
to that of the Neptunes (between 10 and 40 M⊕ ), and dilute the
significance of any trend visible for Neptunes only.
Our results are also in good agreement with the more recent paper
from Wang & Fischer (2015). These authors analysed the Kepler
results and pointed a universal correlation between the occurrence
rate and the host star metallicity. They show that this dependence
is weaker for terrestrial planets than for gas-dwarf planets. This is
consistent with the fact that we can extrapolate such a correlation
for our Neptune sample but not for our super-Earth sample, while
both are roughly equal in size. However we note that they use pho-
tometrically derived KIC metallicities, that have a poor precision.
5 DISCUSSION
While the observed correlation does not imply causation, it is in-
teresting to consider the possibility that the metallicity would drive
the maximal mass of these small planets. A mass-metallicity upper
boundary fits well within the core-accretion theory (Pollack et al.
1996). Indeed, a metal poor star could see its planets forming smaller
cores, that will thus accrete gas less efficiently. This translates as a
maximum mass rather than a strict correlation probably because of
other phenomena, such as the competition between multiple planets
that decrease the quantity of matter available for a single one.
The correlation between this boundary and the period is also
peculiar. A possible explanation would be that the highly irradiated
parts of a protoplanetary disc are more depleted of volatile elements
due to intense radiation pressure. The planets forming in these
regions have a much lower amour of gas and rely more exclusively
on the presence of heavier elements. Such a scenario implies that
these planets are formed in situ, or at least that a significant fraction
of their total mass is acquired during or after the migration. Another
possible explanation is that planets do form at all metallicities, but
that in metal-poor discs Neptune-like planets form farther out or
do not migrate as quickly as in metal-rich discs. These planets
would consequently lie outside the period ranges that are probed by
current surveys. This is the mechanism proposed by Adibekyan et al.
(2013), who showed that the periods of planets more massive than
10 M⊕ orbiting metal-poor stars are preferentially longer than those
orbiting metal-rich stars. However, the role of other parameters like
the mass of the protoplanetary disc, the number of planets and
MNRAS 461, 1841–1849 (2016)
1848 B. Courcol, F. Bouchy and M. Deleuil
architecture of the system or the migration type should be taken
into account.
Similar studies have been performed on the Kepler Objects of
Interest sample when considering the radius instead of the mass.
Buchhave et al. (2014) show that the mean host star metallicity is
statistically higher for larger planets, with spectroscopically derived
metallicities. They categorize them into three groups of increasing
metallicities: the terrestrial planets RP < 1.7R⊕ , gas dwarf planets
1.7R⊕ < Rp < 3.9R⊕ and gas or ice giants 3.9R⊕ < Rp . These
results are consistent with the trend we observe for the masses.
Based on the same metallicity values, Dawson, Chiang & Lee
(2015) go one step further and show that for period greater than
15 d, for which there is no significant photoevaporation, there is a
lack of rocky planets (RP < 1.5R⊕ ) around metal rich star, while
all types of planets are found around metal poor stars. The semi-
empirical mass–radius relation of small rocky exoplanets obtained
by Zeng, Sasselov & Jacobsen (2015) shows that 1.5R⊕ corresponds
to 4.2M⊕ . In our sample, no planets smaller than 4.2 M⊕ at less
than 15 d are found around stars more metallic than the Sun, which
is therefore in agreement with Dawson et al. (2015).
Finally, one additional remark can be made. We know at least 18
planets more massive than 40 M⊕ with [Fe/H] < -0.4 (source: exo-
planets.org), when no Neptunes are found in that metallicity range,
cf Section 4. Moreover the orbital periods of these planets are very
different, from 2.96 d (WASP-98 b) to 956 d (HD 181720 b). There
is therefore a turnover in the distribution that coincides with the
planet desert that separates Neptunian and Jovian planets, between
40 and 60 M⊕ . This implies that the formation mechanism of gi-
ant planets is significantly different to that of low-mass planets and
reinforces the role of metallicity in planetary formation.
6 CONCLUSION
We compiled all the known low-mass planets (Msin(i) < 40M⊕ )
as of 2015 December with a good precision on both minimal mass
(<20 per cent) and host star metallicity (<0.2 dex). We studied the
resulting 157 objects in the mass-metallicity diagram. The maxi-
mum mass at any given metallicity bin increases with metallicity in
a seemingly linear manner (with a possible plateau below −0.5 dex),
with a desert for Neptune like planets with low host-star metallicity.
Observational or selection biases could not reproduce this feature,
which is therefore physical in nature, as they would have either no
impact or favour the detection of more massive planets that would
have been located in the observed desert.
We demonstrated that there is a dependence between this upper
boundary and the planets period. The boundary shrinks at shorter
periods and is more expanded at longer periods. This is in agree-
ment with an in situ core-accretion formation mechanism where
the irradiation depleted more efficiently the protoplanetary disc of
its volatile content. Alternatively, it could mean that planets around
metal-poor stars form at longer periods or migrate not as quickly,
and would orbit outside the period range probed by current sur-
veys. However, a number of parameters were not considered in this
study, such as planet multiplicity, planetary migrations and system
architecture.
This period dependent boundary has implications on the low-
mass planet frequency. The distribution in metallicity of Neptunes
hosts (between 10 and 40 M⊕ ) indicates that their frequency is
likely to increase with metallicity, and that this effect is stronger for
high irradiation planets. On the contrary, no such effect is visible
for the Super-Earth population (<10 M⊕ ). These results, that are
corroborated by recent papers on the Kepler sample (Buchhave et al.
MNRAS 461, 1841–1849 (2016)
2014; Dawson et al. 2015; Wang & Fischer 2015), might change
the claim that the occurrence rate of Neptunes is not correlated
to metallicity, as opposed to giant planets. The discrepancy with
previous studies (Sousa et al. 2008; Ghezzi et al. 2010; Mayor
et al. 2011; Sousa et al. 2011) can be explained by a larger sample
and a distinction between Neptunes and super-Earths, that exhibit
different behaviours.
The statistical properties of low-mass planets are crucial to un-
derstand planetary formation. New properties such as this exclusion
Downloaded from https://academic.oup.com/mnras/article-abstract/461/2/1841/2608353 by The Open University user on 23 April 2020
zone should be now explained in the framework of planet formation
models. It is of prime importance to significantly increase the sam-
ple of planets in this small mass domain with accurate parameters
in order to refine and assess the robustness of the current result. Our
result have important implication regarding the expected results of
radial velocity surveys targeting metal-poor stars. More specifically,
programs dedicated to the observation of metal-poor stars on long
time-scales could confirm if Neptune-like planets do exist at longer
periods. Upcoming programs dedicated to transit search around
bright stars (TESS, CHEOPS, NGTS) alongside ground RV facil-
ities will enable the study of thousands of low-mass planets. Not
only a much larger sample but additional parameters such as stellar
age, multiplicity, true mass, radius or density will provide new and
crucial insights on this exclusion zone, the impact of irradiation,
the differences between Neptunes and super-Earth populations and
planetary formations processes.
AC K N OW L E D G E M E N T S
We sincerely thank Christian Marinoni for useful comments and
conversations. We also gratefully acknowledge the Programme
National de Planétologie (financial support) of CNRS/INSU.
REFERENCES
Adibekyan V. Z. et al., 2013, A&A, 560, A51
Anglada-Escudé G., Tuomi M., 2012, A&A, 548, A58
Anglada-Escudé G. et al., 2013, A&A, 556, A126
Anglada-Escudé G. et al., 2014, MNRAS, 443, L89
Astudillo-Defru N. et al., 2015, A&A, 575, A119
Baglin A., 2003, Adv. Space Res., 31, 345
Bakos G. Á. et al., 2010, ApJ, 710, 1724
Bakos G. Á. et al., 2015, preprint (arXiv:1507.01024)
Bonfils X. et al., 2007, A&A, 474, 293
Bonfils X. et al., 2011, A&A, 528, A111
Bonfils X. et al., 2013, A&A, 556, A110
Borucki W. J., Koch D. G., 2011, Proc. IAU Symp., 276, 34
Borucki W. J. et al., 2010, ApJ, 713, L126
Bouchy F. et al., 2009, A&A, 496, 527
Buchhave L. A. et al., 2014, Nature, 509, 593
Burt J., Vogt S. S., Butler R. P., Hanson R., Meschiari S., Rivera E. J., Henry
G. W., Laughlin G., 2014, ApJ, 789, 114
Butler R. P. et al., 2006, ApJ, 646, 505
Carter J. A., Winn J. N., Holman M. J., Fabrycky D., Berta Z. K., Burke
C. J., Nutzman P., 2011, ApJ, 730, 82
Carter J. A. et al., 2012, Science, 337, 556
Courcol B. et al., 2015, A&A, 581, A38
Dawson R. I., Chiang E., Lee E. J., 2015, MNRAS, 453, 1471
Delfosse X. et al., 2013, A&A, 553, A8
Demory B.-O. et al., 2013, ApJ, 768, 154
Dı́az R. F. et al., 2016, A&A, 585, A134
Dressing C. D. et al., 2015, ApJ, 800, 135
Dumusque X. et al., 2012, Nature, 491, 207
Dumusque X. et al., 2014, ApJ, 789, 154
Endl M. et al., 2012, ApJ, 759, 19
 An upper boundary in the mass-metallicity plane of exo-Neptunes 1849
Fischer D. A. et al., 2012, ApJ, 745, 21 Pepe F. et al., 2007, A&A, 462, 769
Forveille T. et al., 2009, A&A, 493, 645 Pepe F. et al., 2011, A&A, 534, A58
Forveille T. et al., 2011, preprint (arXiv:1109.2505) Pepe F. et al., 2013, Nature, 503, 377
Fulton B. J. et al., 2015, ApJ, 805, 175 Pollack J. B., Hubickyj O., Bodenheimer P., Lissauer J. J., Podolak M.,
Ghezzi L., Cunha K., Smith V. V., de Araújo F. X., Schuler S. C., de la Reza Greenzweig Y., 1996, Icarus, 124, 62
R., 2010, ApJ, 720, 1290 Porto de Mello G. F., da Silva R., da Silva L., de Nader R. V., 2014, A&A,
Gonzalez G., 1997, MNRAS, 285, 403 563, A52
Hadden S., Lithwick Y., 2014, ApJ, 787, 80 Rivera E. J., Laughlin G., Butler R. P., Vogt S. S., Haghighipour N.,
Hartman J. D. et al., 2011, ApJ, 728, 138 Meschiari S., 2010, ApJ, 719, 890
Haywood R. D. et al., 2014, MNRAS, 443, 2517 Robertson P., Mahadevan S., 2014, ApJ, 793, L24

Downloaded from https://academic.oup.com/mnras/article-abstract/461/2/1841/2608353 by The Open University user on 23 April 2020

Hébrard G. et al., 2010, A&A, 512, A46 Sanchis-Ojeda R., Rappaport S., Winn J. N., Levine A., Kotson M. C.,
Hirano T. et al., 2012, ApJ, 759, L36 Latham D. W., Buchhave L. A., 2013, ApJ, 774, 54
Howard A. W. et al., 2009, ApJ, 696, 75 Santos N. C., Israelian G., Mayor M., Bento J. P., Almeida P. C., Sousa
Howard A. W. et al., 2011a, ApJ, 726, 73 S. G., Ecuvillon A., 2005, A&A, 437, 1127
Howard A. W. et al., 2011b, ApJ, 730, 10 Santos N. C. et al., 2013, A&A, 556, A150
Howard A. W. et al., 2014, ApJ, 794, 51 Ségransan D. et al., 2011, A&A, 535, A54
Huber D. et al., 2013, Science, 342, 331 Soubiran C., Le Campion J.-F., Cayrel de Strobel G., Caillo A., 2010, A&A,
Jenkins J. S. et al., 2013, ApJ, 766, 67 515, A111
Johnson J. A., Apps K., 2009, ApJ, 699, 933 Sousa S. G. et al., 2008, A&A, 487, 373
Laws C., Gonzalez G., Walker K. M., Tyagi S., Dodsworth J., Snider K., Sousa S. G., Santos N. C., Israelian G., Mayor M., Udry S., 2011, A&A,
Suntzeff N. B., 2003, AJ, 125, 2664 533, A141
Lissauer J. J. et al., 2013, ApJ, 770, 131 Tinney C. G., Butler R. P., Jones H. R. A., Wittenmyer R. A., O’Toole S.,
Lo Curto G. et al., 2010, A&A, 512, A48 Bailey J., Carter B. D., 2011, ApJ, 727, 103
Lo Curto G. et al., 2013, A&A, 551, A59 Tuomi M., 2014, MNRAS, 440, L1
Lovis C. et al., 2006, Nature, 441, 305 Tuomi M., Anglada-Escudé G., Gerlach E., Jones H. R. A., Reiners A.,
Lovis C. et al., 2011, A&A, 528, A112 Rivera E. J., Vogt S. S., Butler R. P., 2013, A&A, 549, A48
Maness H. L., Marcy G. W., Ford E. B., Hauschildt P. H., Shreve A. T., Basri Valenti J. A. et al., 2009, ApJ, 702, 989
G. B., Butler R. P., Vogt S. S., 2007, PASP, 119, 90 Van Grootel V. et al., 2014, ApJ, 786, 2
Marcy G. W. et al., 2014, ApJS, 210, 20 Vanderburg A. et al., 2015, ApJ, 800, 59
Masuda K., 2014, ApJ, 783, 53 Vogt S. S. et al., 2010, ApJ, 708, 1366
Masuda K., Hirano T., Taruya A., Nagasawa M., Suto Y., 2013, ApJ, 778, Vogt S. S. et al., 2015, ApJ, 814, 12
185 Wang J., Fischer D. A., 2015, AJ, 149, 14
Mayor M. et al., 2009, A&A, 493, 639 Wittenmyer R. A. et al., 2014, ApJ, 791, 114
Mayor M. et al., 2011, preprint (arXiv:1109.2497) Wright J. T., Upadhyay S., Marcy G. W., Fischer D. A., Ford E. B., Johnson
Melo C. et al., 2007, A&A, 467, 721 J. A., 2009, ApJ, 693, 1084
Mordasini C. et al., 2011, A&A, 526, A111 Xie J.-W., 2014, ApJS, 210, 25
Mordasini C., Alibert Y., Georgy C., Dittkrist K.-M., Klahr H., Henning T., Zeng L., Sasselov D., Jacobsen S., 2015, preprint (arXiv:1512.08827)
2012, A&A, 547, A112
Mortier A. et al., 2016, A&A, 585, A135
Motalebi F. et al., 2015, A&A, 584, A72
Neves V. et al., 2012, A&A, 538, A25
O’Toole S. et al., 2009a, ApJ, 697, 1263
O’Toole S. J., Jones H. R. A., Tinney C. G., Butler R. P., Marcy G. W.,
Carter B., Bailey J., Wittenmyer R. A., 2009b, ApJ, 701, 1732 This paper has been typeset from a TEX/LATEX file prepared by the author.

MNRAS 461, 1841–1849 (2016)